Approximate String Matching of Power System Substation Names

When running simulations, we often use power system cases from different organizations. Each organization has its own naming schemes, which usually result in the same bus being numbered and spelled differently for different cases (e.g., through contractions). While humans can easily identify a contraction, the large number of buses requires an automated method, which is difficult. While research exists on approximate string matching, the application to power system bus names reveals unique challenges. This paper proposes an approximate string matching algorithm for bus names. The algorithm uses an optimally chosen string distance measure to quantify the difference between pairs of bus names from two different cases, and an improved Hungarian algorithm to find the overall minimum distance pairing of buses. Through testing, it was found that using the edit distance produced the best results. Using our algorithm, 76% of the matches in a training case were identified, with 99% accuracy. This is a large improvement over the 32% of exact matches.